Getting 1011 Error when i have been using bi-directional websocket with gemini liveAPI. What's the solution?

[bavidoddi-suresh]

I’m consistently hitting a 1011 Internal Error while using bi-directional WebSocket streaming with the Gemini Live API.

Setup details:

Using a sub-sequential agent workflow

Frontend connects via WebSocket

Backend is deployed on Cloud Run

Agent communicates with Gemini through the Live API stream

The issue:
The WebSocket connection drops with a 1011 error, and I’m unable to pinpoint the root cause. No clear logs or explanations from the server side.

Has anyone faced this? What are the common triggers or recommended fixes for 1011 errors with Gemini Live API WebSocket sessions?

[adk-bot]

Response from ADK Answering Agent (experimental, answer may be inaccurate)

TLDR: A WebSocket 1011 error in ADK indicates a generic, unhandled exception on your backend server, not a client-side or network issue. The most likely causes are resource exhaustion (like hitting an API quota for a tool) or using a model that doesn't support the Gemini Live API. The key is to improve your server-side logging to capture the underlying exception that triggers the 1011 error.

---

Hello! Thanks for reaching out. The WebSocket error code 1011 you're seeing is a generic "Internal Error" that our ADK web server uses to close a connection when it encounters an unexpected server-side issue. [1]

This means the problem is likely in your Cloud Run backend, where an unhandled exception occurs during the agent's communication loop. Here are some of the most common triggers for this error:

1. Resource Exhaustion: One of the most common causes is hitting a usage limit for a tool integrated with your agent. For example, internal testing has shown that exceeding the Google Search tool's usage quota will result in a RESOURCE_EXHAUSTED error, which in turn causes the WebSocket to close with a 1011 error. [2]

2. Model Incompatibility: Using a model that does not fully support the Gemini Live API can lead to connection failures. For instance, attempting a streaming session with an incompatible model can cause the connection to be closed unexpectedly. [3]

3. General Server-Side Exceptions: Any unhandled exception within your agent's logic on the backend will cause the server to terminate the WebSocket session with the 1011 error code. The communication is often managed by concurrent tasks, and an exception in any one of them will tear down the entire connection. [4]

To resolve this, I recommend the following troubleshooting steps:

• Inspect Your Cloud Run Logs: This is the most critical step. The 1011 error is a symptom, not the cause. You must look for a full exception traceback in your Cloud Run logs that occurs just before the connection is terminated. This will tell you the root cause of the problem.
• Add Granular Logging: Enhance your backend code to add detailed logging within the WebSocket handling tasks (e.g., client_to_agent_messaging and agent_to_client_messaging). Log the messages being sent and received so you can identify the exact point of failure. [4]
• Verify Your Model: Double-check that the model you have configured in your agent is compatible with bi-directional streaming via the Gemini Live API. [3]
• Check Tool Quotas: If your agent uses any external tools (like Google Search, etc.), review their usage dashboards and limits to ensure you are not hitting any quotas.

By adding more robust try...except blocks and detailed logging around your agent's execution loop, you should be able to catch the specific underlying error that is currently manifesting as the generic 1011 error.

[1] https://github.com/google/adk-python/blob/main/src/google/adk/cli/adk_web_server.py
[2] https://github.com/google/adk-docs/blob/main/examples/python/snippets/streaming/adk-streaming-ws/tests/test_log_20251029_151045.md
[3] https://github.com/google/adk-docs/blob/main/deploy/gke.md
[4] https://github.com/google/docs/blob/main/streaming/custom-streaming-ws.md